[Dev] [OT] dB

Łukasz Stelmach l.stelmach at samsung.com
Tue Oct 22 13:18:46 GMT 2013


It was <2013-10-22 wto 11:40>, when Jussi Laako wrote:
> On 22.10.2013 12:25, Łukasz Stelmach wrote:
>> log(10) = 2.30258509299 (?)
>> log(exp(1)) = 1
>
> log10(10) == 1, decibels always use 10-base logarithm. Of course you
> can also use natural logarithm, but then:
> log10(x) == ln(x)/ln(10)

I always forget it and need to calculate it everytime again and again ;)

> By the way, I recommend using excellent open source software like
> Octave or Scilab. :)

C-x * q is way quicker for even for logarithms ;) (if you know how to
use them). Then there is python, I can recommend it for some non-trivial
array processing. Octave, however a great tool, i would not recommend it
for anything less than fft.

>> Indeed it is -29.473, that is even less than I though.  Why dBFS? There
>> is no absolute Full-Scale in this case. Full-Scale might be the number
>> of of packages available for the distribution rather than installed on
>> your machine.
>
> Full scale is just 100%, IOW factor of 1. Of course you can have also
> positive dBFS for values exceeding 100%. Decibels should never be used
> without reference point, because it is always a relative value.

The way I've been taught full-scale is something absolute. For example
you've got audio signal with 16 bits per sample and you can't have
anything louder than 0 dBFS. You can't have a sample above 65535.

-- 
Łukasz Stelmach
Samsung R&D Institute Poland
Samsung Electronics
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 489 bytes
Desc: not available
URL: <http://lists.tizen.org/pipermail/dev/attachments/20131022/ff181f7e/attachment.sig>


More information about the Dev mailing list